一种基于BP神经网络的字符认识方法

A Letter Character Recognition Method Based on BP Neural Network

讨论BP神经网络的原理及其缺陷和改进方法.在MATLAB环境下,对含噪声文字符进行识别训练.仿真结果表明,网络收敛速度快,识别分类效果好.

This paper discusses the BP neural network, and points out the BP network＇ s disadvantage, and discus- ses how to improve the network. The recognition system based on the BP neural network was tested by a number of cases with disturbing signals of different level. The tests results show that the running speed is fast, and the errors of recognition rate are very low.